Temperature measurement is a concept that covers the set of methods and facilities to obtain the experimental information about the physical parameter: temperature.
Temperature measurements may be classified into "contact" and "contactless" methods. In contact temperature measurements, the sensing element of the measuring device is put in contact with an object to be measured. In this case, the contact must be long enough for thermal equilibrium to be attained between sensing element temperature and the object. This is one of the main conditions for correct temperature measurements.
The physical basis for temperature measurement is represented by thermometric phenomena, which are physical phenomena affecting the temperature dependence of any parameter (electric, frequency, velocity, etc.) that can be easily and uniquely recorded. The following physical phenomena and temperature dependences are the most commonly used as thermometric phenomena:
volume or pressure of an isolated amount of the gas or liquid;
amplitude or spectra of thermal radiation;
thermal noise spectrum;
nuclear quadruple resonance frequency;
nuclear magnetic resonance frequency;
sonic (ultrasonic) propagation velocity in different media;
cathode luminescence radiation;
interference phenomena in anisotropic media;
condensation and crystallization of substances.
Over thirty thermometric phenomena have been used in these situations. In selecting a thermometric phenomenon, the two most important requirements are related to a sharp sensitivity of the chosen parameter to temperature and insensitivity to other physical parameters, for example, pressure indication, field strength, etc. The most fundamental thermometric phenomena are concerned with the ideal gas (Clapeyron–Clausius) equation, thermal radiation (Planck equation), and thermal noise (Nyquist formula). In these phenomena the complete absence of foreign effects is assumed. Owing to this, practical measurements are related to the thermodynamic temperature by an appropriate process of transfer. The procedure of such transfer is formalized by an official document, the International Temperature Scale.
In practice, the thermodynamic temperature scale is transferred by gas thermometers over the range 10 to 1340 K. For higher temperatures, use is made of the absolute black body thermal radiation described by the Planck law
The third fundamental thermometric phenomenon of thermal noise can, in principle, be used without restrictions however, the measuring errors do not allow the noise thermometry to be used in exact metrological studies. However, noise thermometry still finds widespread use in industrial measurements.
Use of the fundamental thermometric phenomena is advantageous, because temperature measurements do not require calibration. The disadvantage is that in these methods, the measurements are tedious and time-consuming. Therefore, for practical temperature measurements, the thermodynamic temperature scale is transferred to a number of reference temperatures corresponding to equilibrium states of vapor, liquid, and solid substances. These are determined with the highest accuracy possible using modern instrumentation. Substances for the reference points are chosen not only from the considerations of the required equilibrium temperature but also from the simplicity of producing and keeping this substance in the pure form. The main reference points are cited in Table 1. (here t. p. means the triple point).
Interpolation over the intermediate reference points is made using the platinum reference resistance thermometer. Intermediate reference points for neon, mercury, indium, bismuth, cadmium, plumbum, and antimony can also be used. In addition, there is a table of the secondary reference points amounting to 50.
At temperatures above the gold point, the complete absolutely black body radiation pyrometer is the interpolation tool. The necessity to average absolute black radiation and identical absolute absorption in a receiver involves many difficulties, which results in a considerable increase of relative errors associated with transferring the thermodynamic temperature scale at temperatures above the "gold point".
The errors in transferring from the thermodynamic to the practical scale are 10−4 to 10−2 K at temperatures up to 500 K; 10−2 to 3 × 10−1 K up to the "gold point". At higher temperatures, the transfer error can be estimated as the square of 0.1 % of the absolute temperature. The above values characterize the most accurate measurements. The tungsten melting point can be measured with a 15 K uncertainty. The value of this temperature (say, 3694 K, 3421°C) is often presented accurately to the last significant figure, but this apparent accuracy disguises the real uncertainty.
In practical temperature measurements, the error exceeds the above limiting values at ±100°C by one order of magnitude and by two orders at higher temperatures.
Contactless temperature measurements are based on the partial or complete intensity of thermal radiation. The Planck equation is applicable only for an absolute black body. For real emitting surfaces, the radiation flux is related to the black body flux at the same temperature by a factor, i.e., the Emissivity. Extensive tables of emissivity are available, covering nearly all measurement conditions. Nevertheless, establishing an accurate value for emissivity is very difficult and leads to considerable errors. Planck's law establishes the generalized relationship between the radiation density, radiation wavelength, and temperature of an emitter, i.e., of an object of measurement. This opens three main possibilities of contactless temperature measurements: radiation, brightness, and color pyrometry.
Radiation pyrometry is based on measuring a total radiation heat flux. Thus, the radiation pyrometer records the temperature of the absolutely black body emitting the same integral flux as the non-black body does and whose temperature is measured. A correction for the integral emissivity is set by a special vernier directly on the device.
Brightness pyrometers are based on comparing the brightness of monochromatic radiation from the object to be measured, and that of the special filament lamp built into the device. Usually, the filament is observed on the object background through a narrowband light filter typically with a 0.65 mm wavelength range. As in the case of radiation pyrometers, it is necessary to allow for the difference of the radiation intensity to that of an absolutely black body using special tables of emissivity at the chosen wavelength (these are different from the normal radiation pyrometer tables). Usually, a digital or arrow device that records the filament lamp current is built into the brightness pyrometer. In some cases, the filament glow is kept constant and the equilibrium is attained by an optical wedge that controls the radiation brightness of an object of measurement. When the brightness of the filament and the object are equal, the filament disappears on the object background and the filament current or the wedge position uniquely determines the object temperature.
In radiation and brightness pyrometers, the main principal error source lies in the necessity to allow for emissivity. At high temperatures, the error due to an incorrect value of emissivity can exceed 10% of the measured value.
The operation of color pyrometers is based on assuming that there is a fixed emissivity irrespective of wavelength. In this case, the need for information about the emissivity can be excluded, if the temperature is judged by the monochromatic radiation density ratio at two fixed wavelengths. As a rule, the emissivity of substances depends on the monochromatic radiation wavelength. This can be overcome by using three or four spectral ratios, for which it is necessary to use, accordingly, three or four monochromatic light filters with sufficient wavelength contrast.
Among the temperature measurement methods the temperature indicators occupy a special place, for which the temperature is indirected by changing a color, structure, shape.
Metals and their alloys possess a sufficient individual constancy of melting points. There are tables of alloy compositions, for which the melting points discretely vary with a step of 5−10°C and which may be used for temperature measurements from 60 to 1000°C.
In different branches of mechanical engineering there are widespread temperature-sensitive dyes, varnishes, fluxes, pencils, pastes, etc., which change their color at fixed temperature values. The number of the manufactured color temperature indicators exceeds 300, and the increment of the fixed temperatures for the change of color amounts to 2−10°C at a ±1°C error.
In commercial plants, frequently there arises the necessity for plasma and flame temperature measurements. These present difficulties due to the fact that flame radiation varies from linear radiation from a set of monochromatic lines to a complete spectrum, approaching absolute black body radiation. In the latter case, the problem is easily solved by using the pyrometric methods described above. In the case of radiation at distinct lines within the spectrum the method of spectral line conversion is often used. Here, a reference emitter at known temperature is used. Following Kirchhoff's law, the medium is permeated by the reference radiation, which is absorbed in the chosen spectral line.
Temperature. Its Measurement and Control in Science and Industry, Rinehold, 1961.